1. Field of the Invention
The present invention relates generally to improving performance in a packet-based network, and more specifically to accurately estimating future interference power in packet-switched wireless networks.
2. Background
A packet-based network, such as the Internet, can be used to transmit data for various applications and devices, including, for example, cellular phones. A packet-switched cellular network is one example of a packet-based network for transmitting data. In such a packet-switched cellular network, a base station can transmit packets of data that can be reassembled into voice and other information over calls utilizing cellular phones. In such an exemplary cellular network, a transmitter or receiver can be either a base station or a terminal, where the base station is the fixed location that transmits a signal to a cellular (i.e., mobile) phone and a terminal is the mobile unit (e.g., the cellular phone). The future generations of wireless networks must accommodate a growing demand for data packet services. High-speed packet services are necessary for wireless data packet communications, such as Internet protocol (IP), which can provide efficient access to remote networks and servers for telecommuters and facilitate wireless multimedia services such as voice, audio, still-image and video.
It is well known that link adaptation (i.e., the choosing of an appropriate modulation level and associated data rate for transmission of the data packets) and power control (i.e., the dynamic setting of transmission power in order to ensure correct reception while minimizing interference to others) can improve performance of such packet-switched cellular networks. The expected performance gain of these techniques requires accurate prediction of future interference power.
More specifically, link adaptation involves choosing an appropriate modulation level (and associated data rate) for a packet transmission, according to the current link condition. When the radio condition is favorable, a complex modulation can be used for transmission to improve network throughput. On the other hand, when the co-channel interference and/or the signal-path gain between the transmitter and receiver are poor, the packet transmission can be adapted using a robust modulation as a way to ensure correct signal reception. The radio link condition can be determined from the estimated signal-to-interference-plus-noise ratio (SINR), which in turn depends on the interference from neighboring cells, the signal-path gain and the transmission power. Results have shown that significant performance gain can be achieved by appropriate link adaptation algorithms.
Known techniques for dynamic transmission power control have been widely studied and practiced to manage interference in cellular radio networks. To meet the need of bursty traffic characteristics in the wireless packet networks, power control techniques have been proposed to track the (co-channel) interference power and signal-path gain separately. According to the two estimated values, transmission power is then adjusted to yield a given SINR. Results have shown that power control can significantly improve performance of the future wireless packet networks. Thus, in order to obtain the expected performance gain by link adaptation and power control, it is important to estimate future interference power accurately.
In traditional cellular networks that are predominantly circuit-switched and used for voice applications, a transmitter usually remains on for a relatively long period of time. Consequently, interference has a very strong temporal correlation, which enables use of a low-pass filter to remove random measurement errors. For this reason, exponential smoothing techniques are commonly used for that type of environment. Such simple filtering, however, is not adequate for wireless packet-switched networks because such networks are based on packet switching in which each transmitter uses an assigned channel to transmit for a relatively short time before the channel is re-assigned to another transmitter. As a result, the temporal correlation of interference is weaker in the packet-switching environment than in the circuit-switched networks.
As an example, the Enhanced Data rates for GSM Evolution (EDGE) system, one of the standardized third generation networks, supports integrated voice and data services utilizing packetized data. Using multiple modulation and coding levels, the EDGE system employs a link-adaptation technique to adapt packet transmission to one of the modulation levels.
In the same EDGE system (and other wireless packet-switched networks), estimating future interference power with measurement errors involves at least two challenging issues. First, interference power is equal to the difference between the total received power and the power of the desired signal. Measuring total received power is relatively easy. While the power of the desired signal can be measured by filtering based on the training symbols for the signal, such measurements can be quite difficult, particularly when the measurement duration is short. The second aspect of the difficulty is that interference measurements typically contain errors (e.g., due to thermal noise).
To illustrate the impact of data packet transmissions that occur in bursts (i.e., transmissions of very short duration), let us consider downlink transmissions in a time division multiple access (TDMA) cellular network with ⅓ frequency reuse. FIG. 1 shows the representative autocorrelation coefficient for the interference power with fixed transmission power, no thermal noise and typical radio parameters. As shown in the figure, depending on the average burst length L, the autocorrelation decreases quickly as a function of the lag time in slots. Although the burst length depends on the data rates and the traffic characteristics of applications, L reaching as low as 10 is common, especially in high-speed networks. Such reduced autocorrelation reveals rapid changes in interference power. As a result, both the interference power and the measurement error now fluctuate from one time slot to the next. Consequently, simple filtering solutions (such as low-pass filters) not only filter out measurement errors, but also smooth out quick changes in interference power, resulting in erroneous estimation of future interference levels.
Co-pending application entitled “A Method and System for Power Control in Wireless Networks Using Interference Prediction with an Error Margin”, U.S. patent application Ser. No. 09/460,993, commonly owned by the assignee hereunder, discloses a method for improving power control using a prediction of future interference power via, in one embodiment, a one-dimensional Kalman filter. The method disclosed, however, does not take advantage of the additional improvement provided by correlating the number of active interferers.
There is, therefore, a need in the art for a method to predict interference power in the presence of measurement errors in packet-switched networks by tracking interference and noise power separately and correlating the number of active interferers.